Map the work before you buy the tool

Most AI projects that stall don't fail because the tool was bad. They stall because nobody agreed which part of the work it was meant to fix. A subscription gets bought, a few people try it, and a few months later it is one more login nobody uses.

The cheapest fix happens before you buy anything: write down how the work actually moves through your business.

Start with one piece of work, end to end

Pick something your team does every week: a client report, a quote, a brief, an onboarding. Write each step as it really happens, not as the process document says it should. Who starts it? Where does the information come from? Who waits on whom?

Look for the breaks, not the busywork

A break is where the work stops or comes back: missing information at a handover, a detail someone has to chase, a review that turns into a rewrite. Breaks cost more than slow steps, because everything after them waits or gets done twice.

Then ask what AI would actually change

For each break, four questions decide whether AI belongs there:

• Would it remove a step, or just make a broken step faster?
• Does it need information you don't capture yet?
• Who checks what it produces, and how long does that take?
• Is anyone allowed to use AI on this material?

Sometimes the honest answer is that the first fix isn't AI at all. A required field on a form or a checklist at handover often comes first, and it is usually what makes AI useful later.

What to do this week

Choose one recurring piece of work, map it with the people who do it, and mark the single break that causes the most rework. That is your starting point, whether you work with us or not.